Announcing our new work: lcm-lookahead.github.io 🥳

We improve IP-Adapter by training with a new identity-lookahead loss and synthetic, consistent data.

The key idea is to propagate image-space losses (ID, CLIP) through LCM-LoRA instead of single-step DDPM approximations.
1/5

Thanks AK for sharing!

TL;DR: InstantStyle is a framework that employs straightforward yet potent techniques for achieving effective disentanglement of style and content from reference images.

Code: github.com/InstantStyle/I…
Project Page: instantstyle.github.io

Oli London This is their story
Beautiful children 😢💔🇮🇱

Excited to share our new work B-LoRA🚀.
With our method, you can use a simplified version of LoRA trained on SDXL to perform style transfer between images and manipulate styles based on text.
Check out our website: B-LoRA.github.io/B-LoRA/

TL;DR: We are looking for Researchers 🌟

We at BRIA AI , having recently secured funding, are set to train 3 foundational models this year (One of them is T2I, with more details to come)

If you wish to train large models 💪but in a startup environment 🚀 - please DM

Snap presents MyVLM

Personalizing VLMs for User-Specific Queries

Recent large-scale vision-language models (VLMs) have demonstrated remarkable capabilities in understanding and generating textual descriptions for visual content. However, these models lack an understanding of

Google presents RT-Sketch

Goal-Conditioned Imitation Learning from Hand-Drawn Sketches

Natural language and images are commonly used as goal representations in goal-conditioned imitation learning (IL). However, natural language can be ambiguous and images can be over-specified.

2024 is the year of text-to-video models 🎥

Join me tomorrow as we dive into ✨Lumiere✨our T2V model from Google AI!
We'll discuss Lumiere's architecture, applications, and more🤩

⏰ tomorrow at 20:00 IDT (10:00 PDT)
🚀 sign up now: docs.google.com/forms/d/e/1FAI…

🚨We uncover a new vulnerability- Pre-Fine-Tuning Weight Recovery
With a few LoRA fine-tuned models we recover the pre-fine-tuning weights🏋️of SoTA models, undoing Stable Diffusion personalization training and Mistral alignment😈
Project: vision.huji.ac.il/spectral_detun…
🧵👇

Introducing Sora, our text-to-video model.

Sora can create videos of up to 60 seconds featuring highly detailed scenes, complex camera motion, and multiple characters with vibrant emotions.

openai.com/sora

Prompt: “Beautiful, snowy…

Happy to share that 🔬ConceptLab🔬 has been accepted to TOG and will be presented in the upcoming #SIGGRAPH2024

Really enjoyed working on this one with Kfir Goldberg and Yuval Alaluf 😊

Code and project page can be found below🧑‍💻

github.com/kfirgoldberg/C…

kfirgoldberg.github.io/ConceptLab/

Excited to share insights about our text-to-video model✨Lumiere✨

Join us at the Vision-Language Club meetup 📅
We'll explore Lumiere's architecture, applications, and results 💡

Big thanks to Dana Arad for the invite!
Secure your spot by signing up👇
forms.gle/WN651zTycsqozn…

We invite you to submit your Motion Generation papers to the HuMoGen #CVPR2024 workshop!
The deadline is on March 19
More details @ humogen.github.io

Video-LaVIT

Unified Video-Language Pre-training with Decoupled Visual-Motional Tokenization

paper page: huggingface.co/papers/2402.03…

In light of recent advances in multimodal Large Language Models (LLMs), there is increasing attention to scaling them from image-text data to more…

Nvidia presents ConsiStory

Training-Free Consistent Text-to-Image Generation

paper page: huggingface.co/papers/2402.03…

enable Stable Diffusion XL (SDXL) to generate consistent subjects across a series of images, without additional training.

Finally 😍 Awesome work!
It's exciting to see vector sketch generation becoming more popular! Endless possibilities! 🖌️

pix2gestalt: Amodal Segmentation by Synthesizing Wholes

paper page: huggingface.co/papers/2401.14…

synthesizes whole objects from only partially visible ones, enabling amodal segmentation, recognition, and 3D reconstruction of occluded objects

This is #Israel .

💙🇮🇱💙
🕊️💙🕊️

#AmYisraelChai
#IStandWithIsrael

Stylized generation 🖌️ is one of my favorite features!
We show that Lumiere can stylize motion🏃‍♀️ by the appearance of a single image 🖼️🎨
using various cool styles from StyleDrop (by Kihyuk Sohn, Nataniel Ruiz et al.).

TLDR: Meet ✨Lumiere✨ our new text-to-video model from Google AI!

Lumiere is designed to create entire clips in just one go!
Seamlessly opening up possibilities for many applications:
Image-to-video 🖼️ Stylized generation 🖌️ Video editing 🪩 and beyond.

See 🧵👇